Nocturia and factors associated with sleep dissatisfaction: a community-based cross-sectional study using explainable machine learning

Abstract Background Nocturia is associated with poor sleep quality, but its relationship with subjective sleep dissatisfaction remains insufficiently understood. In primary care, sleep-related complaints are common, and nocturia may be overlooked as a contributor to sleep dissatisfaction. This study examined the association between nocturia and sleep dissatisfaction and explored age-related differences using explainable machine learning. Methods We conducted a community-based cross-sectional postal survey in 2022 among 10,000 randomly selected residents of Tsukuba City, Japan, aged 45–89 years. A total of 2,148 participants were analyzed. Multivariable logistic regression was used to assess the association between nocturia and sleep dissatisfaction after adjustment for demographic, lifestyle, psychological, and clinical factors. An eXtreme Gradient Boosting (XGBoost) model combined with SHapley Additive exPlanations (SHAP) was used to evaluate the direction and relative magnitude of each variable’s contribution and explore age-related heterogeneity. Results The prevalence of nocturia increased from 8.3% among participants in their 40s to 67.9% among those in their 80s, whereas sleep dissatisfaction peaked at 64.2% among participants in their 50s and did not increase in parallel with age. In multivariable analysis, each additional nocturnal void was independently associated with higher odds of sleep dissatisfaction (odds ratio, 1.61; 95% confidence interval, 1.40–1.85; p < 0.001), together with shorter sleep time, longer sleep onset latency, and lower levels of happiness. SHAP dependence plots suggested age-related patterns in model predictions, prompting exploratory analyses stratified at 65 years. The mean absolute SHAP value for nocturia was numerically higher in the older subgroup model than in the younger subgroup model (0.139 vs. 0.037), although this difference was not formally tested. Conclusions Nocturia was associated with sleep dissatisfaction after adjustment for measured covariates, and exploratory SHAP analyses suggested age-related variation in its predictive importance. In the SHAP dependence plot, the age-related shift toward lower predicted sleep dissatisfaction appeared less pronounced among older adults with more frequent nocturnal urination. Asking about nocturnal voiding frequency may therefore be a feasible and clinically relevant component of sleep evaluation in primary care, particularly in older adults with sleep-related complaints, and may guide further evaluation of urinary symptoms or specialist referral.